Characterized by persistent symptoms following an acute COVID-19 illness, long COVID is a post-viral condition that often limits daily activity and has affected roughly 15% of US adults as of September 2023. [1] Though the pathology of long COVID is not yet understood, several studies have found evidence that COVID-19 vaccination prior to SARS-CoV-2 infection may reduce the risk of long COVID. [2], [3], [4] The magnitude of this reduction, however, is unclear.

There are several reasons for this lack of clarity. First, the definition of long COVID differs between studies, likely because there is not yet a consensus definition. For example, the CDC defines long COVID as persistent symptoms ≥4 weeks following COVID-19 onset. [5] The WHO, however, defines long COVID (“post-COVID condition”) as new or persistent symptoms that: 1) last ≥2 months, 2) occur ≥3 months following onset of COVID-19, and 3) have no alternative explanation. [6] Second, the unvaccinated comparison group differs between studies with regard to timing of COVID-19 onset. In some studies, the unvaccinated group includes only individuals who had COVID-19 during the same timeframe as individuals in the vaccinated group. [7], [8], [9], [10], [11] In other studies, the unvaccinated group includes individuals who had COVID-19 before vaccines became available. [12], [13], [14] These approaches may be susceptible to different types of bias. Third, most studies have used clinical samples from electronic health records (EHRs) [8], [9], [10], [12] or non-probability survey samples. [7], [14] While these studies provide valuable insights, they are susceptible to selection bias (i.e., bias due to the included sample differing from the target population). To reduce this potential bias, studies using more representative samples (e.g., population-based probability samples) are needed.

Thus, the objective of this work was to use a population-based probability sample of adults with COVID-19 to evaluate whether vaccination status at the time of COVID-19 onset was associated with prevalence of long COVID. Further, we considered multiple unvaccinated comparison groups as counterfactuals for the vaccinated group, weighing the strengths and weaknesses of each. We also considered 30-day and 90-day long COVID to incorporate both the CDC and WHO definitions and increase the number of studies with which our work can be compared. We hypothesized that the prevalence of both 30-day and 90-day long COVID would be lower among individuals who completed an initial vaccination series prior to their COVID-19 onset compared to those who were unvaccinated.